• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.1
• /
• pp.11-18
• /
• 2022

De novo drug design is the process of developing new drugs that can interact with biological targets such as protein receptors. Traditional process of de novo drug design consists of drug candidate discovery and drug development, but it requires a long time of more than 10 years to develop a new drug. Deep learning-based methods are being studied to shorten this period and efficiently find chemical compounds for new drug candidates. Many existing deep learning-based drug design models utilize recurrent neural networks to generate a chemical entity represented by SMILES strings, but due to the disadvantages of the recurrent networks, such as slow training speed and poor understanding of complex molecular formula rules, there is room for improvement. To overcome these shortcomings, we propose a deep learning model for SMILES string generation using variational autoencoders with self-attention mechanism. Our proposed model decreased the training time by 1/26 compared to the latest drug design model, as well as generated valid SMILES more effectively.

• Electronics and Telecommunications Trends
• /
• v.38 no.3
• /
• pp.38-46
• /
• 2023

The value of living a long and healthy life without suffering has increased owing to aging populations, transition to welfare societies, and global interest in health deriving from the novel coronavirus disease pandemic. New drug development has gained attention as both a tool to improve the quality of life and high-value market, with blockbuster drugs potentially generating over 10 billion dollars in annual revenue. However, for newly discovered substances to be used as drugs, various properties must be verified over a long period in a time-consuming and costly process. Recently, the development of artificial intelligence technologies, such as deep and reinforcement learning, has led to significant changes in drug development by enabling the effective identification of drug candidates that satisfy desired properties. We explore and discuss trends in artificial intelligence for de novo drug design.

• Bulletin of the Korean Chemical Society
• /
• v.30 no.6
• /
• pp.1313-1316
• /
• 2009

Cdc25 phosphatases have been considered as attractive drug targets for anticancer therapy due to the correlation of their overexpression with a wide variety of cancers. We have been able to identify five novel Cdc25 phosphatase inhibitors with micromolar activity by means of a structure-based de novo design method with a known inhibitor scaffold. Because the newly discovered inhibitors are structurally diverse and have desirable physicochemical properties as a drug candidate, they deserve further investigation as anticancer drugs. The differences in binding modes of the identified inhibitors in the active sites of Cdc25A and B are addressed in detail.

• The Plant Pathology Journal
• /
• v.33 no.6
• /
• pp.529-542
• /
• 2017

Control of plant diseases is largely dependent on use of agrochemicals. However, there are widening gaps between our knowledge on plant diseases gained from genetic/mechanistic studies and rapid translation of the knowledge into target-oriented development of effective agrochemicals. Here we propose that the time is ripe for computer-aided drug discovery/design (CADD) in molecular plant pathology. CADD has played a pivotal role in development of medically important molecules over the last three decades. Now, explosive increase in information on genome sequences and three dimensional structures of biological molecules, in combination with advances in computational and informational technologies, opens up exciting possibilities for application of CADD in discovery and development of agrochemicals. In this review, we outline two categories of the drug discovery strategies: structure- and ligand-based CADD, and relevant computational approaches that are being employed in modern drug discovery. In order to help readers to dive into CADD, we explain concepts of homology modelling, molecular docking, virtual screening, and de novo ligand design in structure-based CADD, and pharmacophore modelling, ligand-based virtual screening, quantitative structure activity relationship modelling and de novo ligand design for ligand-based CADD. We also provide the important resources available to carry out CADD. Finally, we present a case study showing how CADD approach can be implemented in reality for identification of potent chemical compounds against the important plant pathogens, Pseudomonas syringae and Colletotrichum gloeosporioides.

• Bulletin of the Korean Chemical Society
• /
• v.34 no.4
• /
• pp.1051-1054
• /
• 2013

Closo-carborane has been considered as an efficient boron-carrier for boron neutron capture therapy (BNCT) and an attractive surrogate of lipophilic phenyl or cyclohexyl ring in drug design. Despite a great number of carborane-containing ligands have been synthesized and evaluated, molecular modeling studies of carborane ligands with macromolecules have been rarely reported. We herein describe a facile docking and scoring-function strategy of 16 carborane ligands with an estrogen receptor by using the commercial Gaussian, Chem3D Pro and Discovery Studio (DS) computational programs. Docked poses of the carborane ligands in silico exhibited similar binding modes to that of the crystal ligand in the active site of estrogen receptor. Score analysis of the best docked pose for each ligand indicated that the Ligscore1 and the Dockscore have a moderate correlation with in vitro biological activity. This is the first report on the scoring-correlation studies of carborane ligands with macromolecules. The integrated Gaussian-DS approach has a potential application for virtual screening, De novo design, and optimization of carborane ligands in medicinal chemistry.

• Annual Conference of KIPS
• /
• 2021.05a
• /
• pp.343-346
• /
• 2021

약물 디자인이란 단백질과 같은 생물학적 표적에 작용할 수 있는 새로운 약물을 개발하는 과정이다. 전통적인 방법은 탐색과 개발 단계로 구성되어 있으나, 하나의 신약 개발을 위해서는 10 년 이상의 장시간이 요구되기 때문에, 이러한 기간을 단축하기 위한 인공지능 기반의 약물 디자인 방법들이 개발되고 있다. 하지만 많은 심층학습 기반의 약물 디자인 모델들은 RNN 기법을 활용하고 있고, RNN 은 훈련속도가 느리다는 단점이 있기 때문에 개선의 여지가 남아있다. 이런 단점을 극복하기 위해 본 연구는 self-attention 과 variational autoencoder 를 활용한 SMILES 생성 모델을 제안한다. 제안된 모델은 최신 약물 디자인 모델 대비 훈련 시간을 1/36 단축하고, 뿐만 아니라 유효한 SMILES 를 더 많이 생성하는 것을 확인하였다.